Link and Link (essay)

THE LINK MANIFESTO

I've meant to write this for years. It has finally come to pass that I'm in a writing kind of mood and some guys in the catbox were talking about linking so now is as good a time as any. The different types of links and their uses are described pretty well in the documentation. What is not nearly as clear is how one can get the best out of the most commonly used type of link, the hard link. I'll include pipe links in this category because they're a subset of hard links. For the sake of my lazy old fingers, wherever it says only "link" in this text it should be taken to mean "hard link." Ordinarily I'd post a rambling diatribe in the editor logs but this is strictly on the topic of this node.

LINKING 101: INTRODUCTION TO LINKS

What is the policy on linking?

There really has been no stated official policy on linking apart from the "link and link!" commandment. Though it's touted as one of the basic tenets of noding, there's little elaboration upon the fine or even the gross details of how exactly one is supposed to follow that commandment. This node is widely cited as the authoritative source for instructions yet it doesn't say terribly much. It's been over seven years since a staff member (as TallRoo was back then) had something of substance to say about linking, seven years during which the site, its users, and its content have substantially changed.

A key part of E2's philosophy is to link content to other content. This is a constant and we take it as a given. This is meant to say, links within the site and only within the site are considered the single most important part of noding. While some people believe that this self-containment is ill-advised and leads to the site maintaining an insular character on a web that's tending towards increased integration, this is part of the Grand Experiment that E2 has historically been. The pros and cons of this design are not of the present. I may one day post an editor log with some thoughts on the subject of external links.

Addendum: Since the original version of this article, E2 has changed its policy to permit direct links to external URLs.

Links schminks. Why should I bother?

Links are integral to the way E2 works. Apart from the obvious fact that it is a web site and therefore based on hypertext, there are reasons to link within the site, too. Users are expected to link their writeups because linking is fundamental to having a site with integrated content, in which related concepts and topics are connected to each other. This boils down to the cardinal rule that people need to to use hard links and soft links in order to integrate their writeup and the node with the rest of E2's content.

No node is an island. Linking is not a time-consuming hassle. Any monkey with a typewriter can write but it cannot link. Linking on E2 is the difference between writing an article and crafting a writeup.

A well-linked writeup has a beauty of its own. Apart from being aesthetically pleasing it also shows that you have a firm grasp of some of the finer details of noding, and it helps the reader see which other concepts you, the author, find significant and relevant. If you cannot understand links, you also do not and will not understand E2.

Failure to link a writeup will have several consequences: You will hear from a staff member. You will probably hear from non-staffers too. All of them will be telling you to link the bloody writeup. Half of them will point you to this node. If it has other technical flaws besides being unlinked, it may be deleted by an editor. An unlinked writeup will attract downvotes and very few upvotes. So, you ask if you've made it this far through my rambling, how and how much should I link?

Quoting dem bones from The Perfect Node on the subject of soft links: "There is no arbitrary number of soft-links any given writeup should have. When choosing them just remember to soft-link any closely related nodes or concepts within your writeup (for instance, a writeup about David Bowie should probably have hard and soft-links to [writer], [composer], [artist], [rock and roll], [actor] and [GOD]). Once you've covered the basics the rest of the links are up to you - be creative with it: a writeup with clever links is like reading two writeups."* I may not share bones's enthusiasm for Ziggy Stardust but he's right, and this has been the official Word since he wrote it in 2000. Timeless advice from the Man Himself. This advice is good for hard links and soft links alike. *Some links in the quote redacted because even bones was not perfect.

There's a very nice FAQ article about what links do and what each type is good for. Please use it. If you're new to the site, keep it simple. Your linking will improve with time but you should aim to start out with at least adequate linking.

LINKING 102: TOO MUCH OF A GOOD THING

[Link] [and] [link] [and] [link] [and] [link] [and][...]

While writeups with no links tend to be fixed or picked off before they leave New Writeups, the same is not the case for overlinked ones. I'd say that this is because their problem is, though conspicuous, not as obvious, nor is there a culture that reviles them as much as linkless writeups. Some of the worst examples of poor linking come from very respected users of old. I'll use one of them as an example without naming names. I could use any number of writeups, especially older ones, but I picked one at random from the writeup list of one of the most egregious and persistent offenders. I chose to take a look at a factual, biographical writeup about a significant but not major historical figure from the middle of the 20th century. This is a well-researched, well-written, useful writeup. Its grammar and spelling meet E2 standards. The author is a former Content Editor. I myself upvoted this writeup more than two years ago. BUT...

His writeup has 217 hard links. This in a writeup containing seven functional paragraphs and 927 words. Keep in mind that "words" are counted in a programming sense and the actual words of the text are fewer. Some links span more than one word. This writeup effectively has every fourth word linked. Of the 217 links, only three are pipe links. One of the pipe links links back to the node itself. One is rather facetious and irrelevant to the article. The third is the only truly functional pipe link, linking a plural word to a singular node title. The noder's hard links include a mass of links for common words. Among other dictionary words, he links question, much, close, and year, none of which have any significance in conjunction with the node's subject, and none of which are likely to be of use to the reader. Nobody is going to look up the definition of "much" because the word came up in this writeup. He does this multiple times for the same word at different points of the text.

In contrast, the second writeup in the same node, written by another user about two years later, is twice the length of the first one and has 73 hard links. It also employs pipe links more smartly. Lastname is pipelinked to Firstname Lastname, which is where you would expect to find a writeup on the person, whereas the first writeup links directly to Lastname and not to Firstname Lastname wherever it omits the person's first name in the text. Either writeup would have been complete with no more than 50 links, and at least adequate with 20 links. The first writeup is well written and poorly crafted. The second writeup is much better but still far from perfect.

The Other Place

Let's compare this to Wikipedia since Wikipedia uses hard links and pipe links a lot like E2 does. The Wikipedia article on the same person is about 32KB in size, making it almost twice as long as both our writeups put together. The Wikipedia article has 154 links. Wikipedia's article is, in my opinion, also overlinked. It links to many dates like "July 5" for events that aren't historically significant enough to be considered a major event of the 5th of July in history. It links to marginally relevant subjects. They could have made do with fewer than 100 links. Many links there, too, are poorly pipe linked. Fortunately, Wikipedia's wonderfully intuitive forwarding and disambiguation system takes care of that. That's a luxury that we on E2 can presently only dream of.

My main point, though, is that the Wikipedia article has few of the pure junk links. No gerunds, one or two participles (properly forwarded by the system), no conjunctions or articles. The bulk of the links are also of practical use and directly related to the topic. Only about half a dozen links do not point to an existing article or stub. I'm not going to check to see where the links in our writeups lead to without having a few drinks first. (OK, 72 links in the first writeup and 28 in the second writeup lead nowhere. That's 100 blind links in a single node. *hic*) We may attribute some of the difference to the wiki model, where little stylistic nasties can easily be weeded out by the public.

Why was this a problem again and who's working on it?

On E2 writeups have only one author and such missteps can be corrected only by the author or the editorial staff. Which is why so many of them are still there. The truth is that we have tens of thousands of overlinked writeups and probably several million junk links with only so many bodies willing to invest their time and subject themselves to the tedium of dealing with them. The result of redacting links is seen as being largely cosmetic so it's not a priority in anyone's book. Generating the URLs associated with the links does not substantially deprive us of CPU cycles, though we'd probably find something useful to do with the saved processor time. Of course all this leads to the condition perpetuating itself as the style of old writeups serves as a model for newer ones and we end up with several generations of overlinked rot detracting from otherwise viable content.

While not having a direct impact on resource use, blind links and overlinking are a problem when it comes to spiders. Spiders will follow every unique link on a page. If a writeup has 100 unique hard links, the spider will follow at least every one that it does not already have. Should 25 of those links be blind, that's a hit for every single one of them. It's easy to see that a poorly linked writeup can easily generate 25 useless hits. Multiply that by the number of writeups with that sort of linking and you do have a problem. Homenodes are not immune to the bad habit of overlinking. In both homenodes and writeups, pipe links are often improperly used as commentary. This is sometimes witty or funny but does not help our situation.

Overlinked writeups are also visually unattractive. If you're using a standard, default web browser setup, links will be coloured and underlined. A writeup that has a quarter of its text highlighted is, to put it plainly, ugly. The fewer poorly-crafted links you have, the less likely it is that they'll end up pointing to nothingness two or three years later as the staff tweak node titles to best serve the site. All those NFN writeups from 1999 that defined something as "the plural of x?" Gone, thankfully. I killed hundreds of them myself. If you find a node title that's wrong, talk to an admin or post your correction under edit these E2 titles.

LINKING 201: THE TAO OF LINKING

So how can we improve links?

Here's how everyone can improve their links and with it the E2 experience for users and guests alike:

• Link terms relevant to what you're talking about
• Link to singular nouns. Plural node titles, unless they have a good reason to live or can convincingly plead artistic licence, will die
• Don't link dictionary words unless they're significant in relation to your topic
• Don't link conjunctions. Come on, I should not have to tell anyone that [and] is about as useless a hard link as there could ever be. Yet, it has been done many times
• Don't link back to the writeup itself. Courtesy of the department of redundancy department
• Do not use links for emphasis. Use proper mark-up
• Don't link the same thing over and over. If there are multiple instances of a significant term, link once every few paragraphs or once per screenful of text
• Do link. If your writeup has nothing to link to, consider the possibility that it may not be E2 material
• Don't link for the sake of having links. Don't panic if you have a linkless paragraph
• If in doubt, refer to pick titles carefully to see which variant of a title you ought to link to. If still in doubt, consult an editor
• Britnoders, Aussies and other alternative spellers: See the previous point. E2 titles for all non-literary nodes are standardised on American spelling. I find it as painful as you do but we need a standard. This is not optional. Deal with it and pipe link
• If you feel a need to link creatively, be discreet and discriminating. Link to existing nodes if you can. See what people like TheDeadGuy. Chras4, and oenone do in their writeups
• Creative links that don't exist should be able to stand as a useful node title or be likely to inspire someone to fill them if nodeshelled
• The visible text of a writeup should be meaningful by itself. Don't depend on pipe links to convey its meaning
• If you must link punctuation marks for commentary, link only to existing nodes
• Link view is a marvellous tool for managing your hard links. Use it liberally. Reduce the blind links by nodeshelling what needs to be nodeshelled
• Use external links sparingly and only where really pertinent.

It would be nice if, as a site, we had the means to redo the linking system but it is not a priority. I'd love to see a system that would let you do nifty things like link directly to writeups or users but that's not of the near future. It comes down to user behaviour, where the truth is that the management can't do a whole lot about how people link except try to educate the userbase and give the staff licence to treat links as non-text and therefore editable. It should be stated policy that ALL YOUR [s and ]s ARE BELONG TO US and we will take the liberty to edit out the crappy ones. You own the text but, because the means of linking (like the code that parses brackets as link openers and closers) are technical aids that are provided by the site, E2 maintains some rights over them, especially insofar as they are capable of (individually or in bulk) having an impact upon the site's operation, image, and performance. Even then, as things are today we'd still be trying to swim against a current. Nothing that we could decree or code will work until we put an end to the culture of overlinking.

Link less and prosper.

Comments? /msg me. I'm sure the list could use a few refinements.